Re: openat2.h location (openat2.2)

[Florian Weimer <fweimer@redhat.com>]

* Aleksa Sarai:

> On 2020-06-30, Michael Kerrisk (man-pages) <mtk.manpages@gmail.com> wrote:
>> [CC += Aleksa, linux-man@]
>> 
>> Hello Petr,
>> 
>> I'm bringing Aleksa, the creator of the system call, into CC.
>> 
>> Aleksa, the issue is that the header file (<openat2.h>) documented in
>> the manual page that you sent does not (yet) exist (and the bug
>> reporter therefore suggests that the manual page should say to include
>> <linux/openat2.h>.
>> 
>> I'm guessing that there is (or was) a plan to add a <openat2.h> header
>> file to glibc. What is the status?
>
> Yes, Florian asked me to put the openat2 definitions into a separate
> header (openat2.h) so that it could be just copied by glibc rather than
> having to do any more hacks around <fcntl.h>.

We still need to document the original *at functions. 8-( That's a
precondition for adding openat2 (which should come with its own
documentation right from the start).

<https://sourceware.org/pipermail/libc-alpha/2020-May/113545.html>

> I guess those plans haven't come to fruition yet? I'm fine with telling
> users to use <linux/openat2.h> in the meantime.

If we include <linux/openat2.h> from the glibc header, the transition
should be unproblematic because there won't be any multi-definition
errors.

Thanks,
Florian